独立分量分析在混沌信号分析中的应用

Independent Component Analysis in Chaotic Signal Analysis

本文从高阶统计信息角度来研究混沌信号 ,利用独立分量分析 (IndependentComponentAnalysis以下简称ICA)来探求混沌信号的特征结构 .集中讨论了两个问题 .一是基于ICA及相关的近似熵提出了一种新的方法来解决相空间重构中最佳延迟的选择问题 ;二是利用ICA来实现局部本征维数 (LID)求取算法 ,从而对混沌吸引子的维数进行估计 .实验结果表明该方法能正确分别混沌信号与非混沌信号 ,尤其是能正确判别 1/f随机过程不是混沌过程 .

The higher order statistics information of chaotic system was explored, and a new blind signal separation method was presented, which is called independent component analysis and can be used for chaotic signal processing. A new method to search the optimum delay of phase space reconstruction was proposed, then the local intrinsic dimension from independent component analysis was estimated. The corresponding experiments show, that the latter can distinguish between a 1/f process and a chaotic system.